Sustainability tracking using CloudWatch

While CloudWatch does not natively track carbon emissions directly, it plays a crucial role in collecting granular, near real-time resource utilization data—such as CPU usage, storage activity, and network throughput—which are key inputs for estimating energy consumption and, by extension, carbon footprint.

You can use CloudWatch to:

• Track resource utilization: Monitor EC2, Lambda, S3, and other services for usage spikes, idle resources, or inefficient scaling.
• Set up alarms and automation: Trigger actions (e.g., decommissioning idle instances or rightsizing resources) when thresholds are exceeded, reducing unnecessary energy consumption.
• Feed data to carbon estimation tools: Export CloudWatch metrics to external tools or dashboards that estimate emissions based on usage patterns.

Ideallly you should use AWS Customer Carbon Footprint Tool (CCFT), which estimates the carbon emissions associated with your AWS workloads. However, CCFT data is updated monthly with a three-month lag and is not available in real time. To bridge this gap, AWS recommends using sustainability proxy metrics—such as compute hours, storage usage, and data transfer—tracked in near real time via CloudWatch. These proxy metrics serve as actionable indicators for teams aiming to optimize workloads for sustainability.

As per WAF,

Here are the opportunities for Optimization:-

• Continuous Improvement: Real-time proxy metrics enable rapid feedback loops for optimization—teams can experiment with workload changes and immediately observe the impact on resource usage.
• Automated Remediation: CloudWatch alarms can trigger automated workflows (e.g., scaling down resources or pausing non-critical workloads during off-peak hours) to reduce energy use.
• Alignment with Sustainability Pillar: The AWS Well-Architected Framework’s Sustainability Pillar encourages organizations to architect for minimal resource usage and maximize efficiency. Using CloudWatch metrics aligns with best practices such as right-sizing, optimizing storage, and reducing idle resources.

and Limitations are:-

• No Real-Time Carbon Data: CCFT provides only monthly estimates with a lag; real-time carbon emissions are not directly available from AWS.
• Granularity Constraints: CloudWatch metrics are granular for resource usage, but translating these to accurate carbon emissions requires assumptions about underlying infrastructure and regional energy mix, which may not precisely reflect actual emissions.
• Coverage Gaps: Some resources (e.g., serverless architectures or managed services) may not expose all relevant metrics needed for detailed carbon estimation.
• Metric Limits: CloudWatch has query and data retention limits, which may restrict large-scale or highly detailed monitoring setups.